EvoFA: Evolvable Fast Adaptation for EEG Emotion Recognition

TL;DR

EvoFA organically integrates the rapid adaptation of Few-Shot Learning and the distribution matching of Domain Adaptation through a two-stage generalization process, and achieves significant improvements compared to the basic FSL method and previous online methods.

Abstract

Electroencephalography (EEG)-based emotion recognition has gained significant traction due to its accuracy and objectivity. However, the non-stationary nature of EEG signals leads to distribution drift over time, causing severe performance degradation when the model is reused. While numerous domain adaptation (DA) approaches have been proposed in recent years to address this issue, their reliance on large amounts of target data for calibration restricts them to offline scenarios, rendering them unsuitable for real-time applications. To address this challenge, this paper proposes Evolvable Fast Adaptation (EvoFA), an online adaptive framework tailored for EEG data. EvoFA organically integrates the rapid adaptation of Few-Shot Learning (FSL) and the distribution matching of Domain Adaptation (DA) through a two-stage generalization process. During the training phase, a robust base meta-learning model is constructed for strong generalization. In the testing phase, a designed evolvable meta-adaptation module iteratively aligns the marginal distribution of target (testing) data with the evolving source (training) data within a model-agnostic meta-learning framework, enabling the model to learn the evolving trends of testing data relative to training data and improving online testing performance. Experimental results demonstrate that EvoFA achieves significant improvements compared to the basic FSL method and previous online methods. The introduction of EvoFA paves the way for broader adoption of EEG-based emotion recognition in real-world applications. Our code will be released upon publication.

Figures (4)

• Figure 1: Approaches for adapting EEG emotion recognition models. (a) Calibration-based (supervised) methods employ a significant quantity of labeled calibration data in calibration experiments to retrain or fine-tune the model. (b) Offline adaptation methods utilize all target data, encompassing both calibration and test data, to conduct transductive adaptation. (c) Online adaptation approaches prioritize rapid and efficient model adjustments using a smaller amount of calibration data for inductive adaptation, testing on unseen data.
• Figure 2: A schematic representation of the two-step generalization in EvoFA. (a) First generalization (meta-training): initializing an emotion recognition foundation model with strong generalization capabilities based on FSL principles and updating all its parameters to learn these capabilities. (b) Second generalization (fast adapt meta-testing): sample data chronologically from the training dataset to simulate the domain drift of subject data over time. In the inner loop, sequentially reduced the distance between the test data and the support set of the training data. In the outer loop, the model selected an appropriate direction for updating the adapter's parameters. By narrowing the gap with the evolving source, the impact of domain drift on test results can be mitigated.
• Figure 3: t-SNE visualization of emotion recognition output using (a) supervised learning model (GCN) and (b) online calibration model (PN+EvoFA). The different numerical labels correspond to the following emotions: 0-negative, 1-neutral; 2-postive.
• Figure 4: Experimental results of five online calibration models with different amounts of calibration data.